-
Edinburgh Research Explorer
Cooperative secretions facilitate host range expansion
inbacteria
Citation for published version:McNally, L, Viana, M & Brown,
SP 2014, 'Cooperative secretions facilitate host range expansion in
bacteria',Nature Communications, vol. 5, 4594.
https://doi.org/10.1038/ncomms5594
Digital Object Identifier (DOI):10.1038/ncomms5594
Link:Link to publication record in Edinburgh Research
Explorer
Document Version:Publisher's PDF, also known as Version of
record
Published In:Nature Communications
Publisher Rights Statement:This work is licensed under a
Creative Commons Attribution 4.0International License. The images
or other third party material in thisarticle are included in the
article’s Creative Commons license, unless indicated otherwise in
the credit line; if thematerial is not included under the Creative
Commons license, users will need to obtain permission from
thelicense holder to reproduce the material. To view a copy of this
license, visithttp://creativecommons.org/licenses/by/4.0/
General rightsCopyright for the publications made accessible via
the Edinburgh Research Explorer is retained by the author(s)and /
or other copyright owners and it is a condition of accessing these
publications that users recognise andabide by the legal
requirements associated with these rights.
Take down policyThe University of Edinburgh has made every
reasonable effort to ensure that Edinburgh Research Explorercontent
complies with UK legislation. If you believe that the public
display of this file breaches copyright pleasecontact
[email protected] providing details, and we will remove access to
the work immediately andinvestigate your claim.
Download date: 16. Jun. 2021
https://doi.org/10.1038/ncomms5594https://doi.org/10.1038/ncomms5594https://www-ed.elsevierpure.com/en/publications/3a315c57-22fb-4c6f-9c79-011e13b980ab
-
ARTICLE
Received 14 Mar 2014 | Accepted 4 Jul 2014 | Published 5 Aug
2014
Cooperative secretions facilitate host rangeexpansion in
bacteriaLuke McNally1,2, Mafalda Viana3 & Sam P. Brown1,2
The majority of emergent human pathogens are zoonotic in origin,
that is, they can transmit
to humans from other animals. Understanding the factors
underlying the evolution of
pathogen host range is therefore of critical importance in
protecting human health. There are
two main evolutionary routes to generalism: organisms can
tolerate multiple environments or
they can modify their environments to forms to which they are
adapted. Here we use a
combination of theory and a phylogenetic comparative analysis of
191 pathogenic bacterial
species to show that bacteria use cooperative secretions that
modify their environment to
extend their host range and infect multiple host species. Our
results suggest that cooperative
secretions are key determinants of host range in bacteria, and
that monitoring for the
acquisition of secreted proteins by horizontal gene transfer can
help predict emerging
zoonoses.
DOI: 10.1038/ncomms5594 OPEN
1 Centre for Immunity, Infection and Evolution, School of
Biological Sciences, University of Edinburgh, Ashworth
Laboratories, West Mains Road, EdinburghEH9 3JT, UK. 2 Institute of
Evolutionary Biology, School of Biological Sciences, University of
Edinburgh, Ashworth Laboratories, West Mains Road, EdinburghEH9
3JT, UK. 3 Institute of Biodiversity, Animal Health and Comparative
Medicine, Graham Kerr Building, University of Glasgow, Glasgow G12
8QQ, UK.Correspondence and requests for materials should be
addressed to L.M. (email: [email protected]).
NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications 1
& 2014 Macmillan Publishers Limited. All rights
reserved.
mailto:[email protected]://www.nature.com/naturecommunications
-
Predicting the emergence of human pathogens is of
obviousimportance because of both their huge burden on humanhealth
and economic cost1–3. The majority of these
emerging pathogens are zoonotic, that is, they can
transmitbetween humans and animals4,5. Although some
environmentaldrivers of zoonosis have been identified, such as
populationdensity and wildlife biodiversity5, the mechanisms by
whichpathogens extend their host range and become generalists
arepoorly understood6,7.
Organisms can achieve generalism by increasing theirphenotypic
repertoire (for example, by plastically respondingto different
conditions with different behaviours or the activationof different
metabolic pathways), thus becoming tolerantof a wider range of
conditions8. However, organisms can alsoachieve generalism by
modifying the distinct environments theyencounter9,10 so they
resemble a common state to which they arespecialized11,12, in a
process often termed ‘environmentalmodification’9. Bacteria modify
their environments in manyways, most notably via the secretion of
metabolically costlyproteins and metabolites, many of which are
known to beimportant virulence factors13,14. Examples include the
secretionof toxins that kill competitors15–17, digestive exoenzymes
thatmodify the nutrient environment18,19 and biofilms that
protectbacteria from undesirable environments and/or
smothercompetitors20,21. By modifying the local environment,
thesesecretions may not only increase the growth of the
strainsproducing them, but also create an environment to
whichcompetitors are maladapted. Owing to their extracellular
nature,these traits are typically public goods, and have often been
studiedin terms of their social evolutionary dynamics13. However,
theirrole in the evolution of niche breadth remains unexplored.
Herewe show that these secretions allow pathogenic bacteria to
modifyand standardize diverse host environments, thus allowing them
toexpand their host range (Fig. 1).
ResultsComparative analysis. How can we distinguish between
thestrategies of environmental modification via secretions
andclassical generalism in bacteria? Previous work has
suggestedthat bacteria using a classical generalist strategy will
havelarger genomes than specialists to deal with multiple
distinct
environments22. For example, classical generalists may
evolveadditional metabolic pathways to deal with differing
nutrientenvironments. This leads to the prediction that, if
classicalgeneralism is the strategy used by bacteria to extend
their hostrange, the ability to infect multiple hosts will be
positivelycorrelated with genome size. However, if bacteria use a
strategy ofmodifying host environments via secretions we expect a
differentgenomic signature. First, we predict that, if bacteria use
thisenvironmental modification strategy, the ability to infect
multiplehosts will be positively correlated with the number of
secretionscoded in bacterial genomes. The logic for this prediction
is that agreater number of secretions coded in the genome will
allowbacteria to modify host environments to a greater extent
(forexample, digestively simplifying nutrient conditions or
toxifyingthe environment for resident competitors). Second, we
predictthat, if bacteria use this environmental modification
strategy, theability to infect multiple hosts will be negatively
correlated withgenome size. The logic for this prediction is that
investment inmodifying and standardizing the external environment
leads to areduction in the requirement for diverse and specific
geneticadaptations to multiple distinct environments. We
thereforeexpect that the ability of bacteria to infect multiple
host species ispositively correlated with secretome size and
negatively correlatedwith genome size if environmental modification
via secretions isthe major route to host generalism, while we
expect it to bepositively correlated with genome size if bacteria
use a classicalgeneralist strategy.
On the basis of these predictions, we used a
phylogeneticcomparative analysis23 to test whether pathogenic
bacteria useenvironmental modification via secretions to achieve
hostgeneralism (Fig. 2 and Supplementary Fig. 1). We gathered
dataon whether bacteria that infect humans are zoonotic (that
is,infect hosts other than humans) from a previous compilation4.We
also gathered data on bacterial genome sizes and measuredinvestment
in secretions by computational prediction of theirsecretome (that
is, the secreted proteome) sizes from thePSORTdb database24. The
secretome size of bacteria indicatesthe diversity of secreted
proteins that they can use to modifytheir environment, thus
measuring their potential to modifydistinct host environments. In
total, genome sequences andepidemiological data were available for
191 human pathogenspecies (121 zoonotic species, 70 azoonotic
species). As data for
Intra-speciestransmission
Inter-speciestransmission
Intra-speciestransmission
Modification of diseasesite via secretions
Modification of diseasesite via secretions
Specialist onspecies 1
Specialist onspecies 2
Generalist
Environmentalmodifier
Secretions
Figure 1 | Environment-modifying secretions as a route to host
generalism. We consider a scenario where pathogens can potentially
transmit both within
and among host species. Whereas specialists match their hosts
closely (matching colours), generalists that infect multiple hosts
are expected
to have intermediate phenotypes (intermediate colour), meaning
that they will lose to specialists during co-infections. While
environmental modifiers may
lose to specialists and generalists in the unmodified disease
site, they can potentially invade by modifying this environment
(transitions from red/blue to
yellow) via the production of costly secretions (green
triangles) that simplify the environment (loss of patterns).
Specialists and classical generalists
are not adapted to this modified environment, leading to their
exclusion. While specialists and classical generalists are expected
to show complex
adaptations to their host(s) (complex shapes), environmental
modifiers are expected to show simpler adaptations (simple shape),
instead relying on
secretions that modify and simplify their environment.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594
2 NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
different bacterial species are non-independent owing to
theirshared evolutionary history, we used the
whole-genome-basedSUPERFAMILY phylogeny25 to account for common
ancestryamong species. We analysed our data using a
Bayesianphylogenetic mixed model (BPMM), with the zoonotic status
ofeach species as a binary response variable, genome and
secretomesize as predictors, and the phylogenetic relationships
amongspecies as a random effect.
Consistent with the hypothesis of environmental modificationvia
secretions, we found that larger secretome sizes are associatedwith
a higher probability that a pathogen is zoonotic (Fig. 3,BPMM:
parameter estimate (b)¼ 3.23� 10� 2, 95% credibleinterval (CI)¼
5.54� 10� 3 to 6.08� 10� 2). Also in accordance
with the hypothesis of environmental modification via
secretions,but counter to the alternative classical generalism
model22, wefound that genome size had a negative effect on the
probabilitythat a pathogen is zoonotic (BPMM: b¼ � 4.61� 10� 4,
95%CI¼ � 1.59� 10� 5 to � 9.36� 10� 4). These results suggestthat
cooperative environmental modification is the major route tohost
generalism in pathogenic bacteria.
Theoretical model. Why do bacteria use environmental
modi-fication via secretions to achieve generalism instead of the
clas-sical mechanism of increasing their phenotypic repertoire?
Wenow theoretically examine modification of the host
environment
Acinetobacter baumanniiAcinetobacter calcoa
ceticusMoraxella catarrhalisP
seudomonas fluoresc
ensPseudomonas pu
tidaPseudomonas
stutzeri
Pseudomona
s aeruginosa
Salmonella
bongori
Salmonell
a typhimu
rium
Salmonel
la typhi
Shigella
dysenter
iae
Shigella
boydii
Shigell
a flexne
ri
Shigel
la sonn
ei
Escher
ichia c
oli
Citrob
acter
rodent
ium
Citrob
acter
koseri
Klebs
iella p
neum
oniae
Klebs
iella
oxyto
ca
Enter
obac
ter ae
rogen
es
Enter
obac
ter as
buria
e
Enter
obac
ter cl
oaca
e
Cron
obac
ter sa
kaza
kii
Rahn
ella
aqua
tilis
Edwa
rdsie
lla ta
rda
Prov
iden
cia s
tuar
tii
Prot
eus
mira
bilis
Yers
inia
pse
udot
uber
culo
sis
Yers
inia
pes
tis
Yers
inia
ent
eroc
oliti
ca
Serra
tia p
rote
amac
ulan
s
Hae
mop
hilu
s pa
rain
fluen
zae
Hae
mop
hilu
s in
fluen
zae
Hae
mop
hilu
s du
crey
i
Pas
teur
ella
mul
toci
da
Act
inob
acill
us p
leur
opne
umon
iae
Vib
rio v
ulni
ficus
Vib
rio c
hole
rae
Vib
rio fu
rnis
sii
Vib
rio p
arah
aem
olyt
icus
Aer
omon
as v
eron
ii
Aer
omon
as h
ydro
phila
Fra
ncis
ella
tula
rens
is
Legi
onel
la lo
ngbe
acha
e
Legi
onel
la p
neum
ophi
la
Cox
iella
bur
netii
susodon retcabolehciDS
teno
trop
hom
onas
mal
toph
ilia
Ralstonia pickettii
Burkholderia cepacia
Burkholderia pseudom
allei
Burkholderia m
allei
Achrom
obacter xylosoxidans
Bordetella parapertussis
Bordetella pertussis
Bordetella bronchiseptica
Bordetella avium
Com
amonas testosteroni
Delftia acidovorans
Chrom
obacterium violaceum
Neisseria m
eningitidis
Neisseria gonorrhoeae
Neisseria lactam
ica
Ehrlichia chaffeensis
Orientia tsutsugam
ushi
Rickettsia australis
Rickettsia akari
Rickettsia massiliae
Rickettsia japonica
Rickettsia africae
Rickettsia conorii
Rickettsia rickettsii
Rickettsia felis
Rickettsia typhi
Rickettsia prowazekii
Ochrobactrum anthropi
Brucella melitensis
Bartonella quintana
Bartonella bacilliformis
Bartonella henselae
Arcobacter butzleri
Campylobacter lari
Campylobacter jejuni
Campylobacter fetus
Campylobacter curvus
Campylobacter concisusHelicobacter pylori
Helicobacter cinaedi
Wolinella succinogenesStreptobacillus moniliformisLeptotrichia
buccalisSebaldella termitidisFusobacterium nucleatumUreaplasma
urealyticumMycoplasma genitaliumMycoplasma pneumoniae
Mycoplasma hominis
Mycoplasma fermentans
Enterococcus hirae
Enterococcus faeciu
m
Enterococcus fae
calis
Streptococcus
equi
Streptococc
us pyogenes
Streptococ
cus uberis
Streptococ
cus agalac
tiae
Streptoc
occus int
ermedius
Streptoc
occus go
rdonii
Streptoc
occus p
neumon
iae
Streptoc
occus m
itis
Strepto
coccus
suis
Strept
ococcu
s muta
ns
Strept
ococcu
s saliv
arius
Lysinib
acillus
spha
ericus
Bacill
us pu
milus
Bacill
us lic
henif
ormis
Bacill
us su
btilis
Bacil
lus th
uring
iensis
Bacil
lus ce
reus
Bacil
lus an
thrac
is
Bacil
lus co
agula
ns
Liste
ria se
elige
ri
List
eria
ivan
ovii
List
eria
mon
ocyt
ogen
es
List
eria
wel
shim
eri
Stap
hylo
cocc
us lu
gdun
ensi
s
Stap
hylo
cocc
us e
pide
rmid
is
Stap
hylo
cocc
us a
ureu
s
Stap
hylo
cocc
us s
apro
phyt
icus
Sta
phyl
ococ
cus
haem
olyt
icus
Bre
viba
cillu
s br
evis
Clo
strid
ium
diff
icile
Filif
acto
r alo
cis
Eub
acte
rium
lim
osum
Eub
acte
rium
rec
tale
Clo
strid
ium
nov
yiC
lost
ridiu
m te
tani
Clo
strid
ium
per
frin
gens
Clo
strid
ium
bot
ulin
umF
ineg
oldi
a m
agna
Aci
dam
inoc
occu
s fe
rmen
tans
Vei
llone
lla p
arvu
laE
rysi
pelo
thrix
rhu
siop
athi
aeLe
ptos
pira
bor
gpet
erse
nii Leptospira interrogans
Treponem
a pallidumB
orrelia recurrentisB
orrelia crociduraeB
orrelia duttoniiB
orrelia turicataeB
orrelia hermsii
Borrelia burgdorferi
Gardnerella vaginalis
Bifidobacterium
dentium
Rothia dentocariosa
Arcanobacterium
haemolyticum
Propionibacterium
acnes
Rhodococcus equi
Rhodococcus erythropolis
Nocardia farcinica
Gordonia bronchialis
Mycobacterium
leprae
Mycobacterium
africanum
Mycobacterium
bovis
Mycobacterium
tuberculosis
Mycobacterium
ulcerans
Mycobacterium
marinum
Mycobacterium
avium
Mycobacterium
smegm
atis
Mycobacterium abscessus
Tsukamurella paurometabola
Corynebacterium ulcerans
Corynebacterium pseudotuberculosis
Corynebacterium diphtheriae
Corynebacterium urealyticum
Corynebacterium jeikeium
Saccharomonospora viridis
Eggerthella lenta
Bacteroides thetaiotaomicron
Bacteroides vulgatus
Bacteroides fragilis
Prevotella melaninogenica
Prevotella denticola
Prevotella intermedia
Prevotella ruminicola
Porphyromonas asaccharolytica
Porphyromonas gingivalis
Capnocytophaga canimorsus
Capnocytophaga ochracea
Chlamydophila pneumoniae
Chlamydia trachomatis
Chlamydophila psittaci
AzoonoticZoonoticSmall secretomeLarge secretomeSmall genomeLarge
genome
Figure 2 | The phylogenetic distribution of zoonosis, genome
size and secretome size. The phylogenetic distribution of zoonotic
status, secretome size
and genome size is shown. Large genomes and secretomes are those
greater than the median and small are less than or equal to the
median.
Note that the tree is ultrametricized for illustrative purposes
only.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594 ARTICLE
NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications 3
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
as a strategy to achieve generalism under a simple nested
epide-miological scenario. Our model focuses on the
epidemiologicalconsequences of secretions that modify a strain’s
environmentrather than the conditions for the initial evolution of
thesesecretions, which are already well understood13. We use
asusceptible–infected–susceptible epidemiological model,
withexplicit within-host dynamics governed by the
replicatorequation26, to model the dynamics of competing
pathogenstrains. We consider a scenario where pathogens can
potentiallyinfect two different host species, and where
transmission betweenthese species is possible. We consider four
different strain types:two specialist strains that each infect one
of the two host species,classical generalists that can infect both
species andenvironmental modifiers that can infect both species
byinvestment in cooperative modification of the hostenvironments
into a common simplified state. We make fourkey assumptions in our
model. First, we assume that a generalist’sgrowth rate g is lower
than the growth rate s of a specialist withinits preferred host
(gos), that is, that there is a trade-off in theevolution of
classical generalism8. We further assume thatenvironmental
modifiers have a growth rate bEM� c in a host,where b is the
benefit from growing in the modified hostenvironment that they
create, c (cob) is the cost of investment insecretions to modify
the host environment and EM is thefrequency of the environmental
modifier strain within the host.The growth rate of environmental
modifiers is therefore positivelyfrequency dependent (that is,
increasing with EM) asmodification of the host environment is a
collective endeavour.This modification of conditions within the
disease site ispredicted to reduce the growth rates of specialists
and classicalgeneralists during co-infections as it creates an
environment towhich they are maladapted (for example, by modifying
thenutrient environment and/or community composition to a newstate
in which specialists growth rate will be reduced). We modelthis
effect by setting the growth rates of specialists and
classicalgeneralists during co-infection with environmental
modifiersas s(1�EM) and g(1�EM), respectively. Finally, we
assumethat environmental modifiers’ growth rate in a single
straininfection is lower than that of a specialist in their
preferred hostspecies (b� cos).
Our theoretical model shows that strains using a strategy
ofenvironmental modification via cooperative secretions can
invadepopulations of specialist pathogens under a wider range
ofconditions than classical generalists (Fig. 4). Both
classicalgeneralists and environmental modifiers are favoured by
higher
contact rates between host species (favouring generalism)
andhigher clearance rates of infection (reducing competition
withspecialists). However, the condition for environmental
modifiersto invade the specialist population is less stringent than
thecondition for classical generalists (that is, their basic
reproductivenumber is always greater). This occurs because of what
we refer toas a ‘scorched earth’ effect. Environmental modifiers
alter the hostenvironment, increasing their own growth rate, while
alsoreducing the growth rate of any co-infecting specialist, which
isnot adapted to this modified environment. This means that,
evenwhen they have a lower growth rate in single strain
infections,environmental modifiers can compete successfully
againstspecialists within a host by sufficiently reducing the
specialist’sgrowth rate relative to their own. We also note that,
while wehave assumed that modification of the host
environmentincreases the growth rate of environmental modifiers, in
principlethis same effect may occur when environmental
modifierssecretions toxify the environment for themselves also, so
longas they reduce the growth rate of specialists to a greater
extent12.
DiscussionOur results have major implications for our
understanding of theconsequences, and evolutionary function, of
bacterial sociality.Secretions are generally considered to be
social traits in bacteria:they will either help or harm the
surrounding cells13,14. The greatabundance of these social
secretions has led to a wealth ofliterature exploring selective
forces governing the evolution ofbacterial sociality13,27. Our
results show that one of the majorconsequences of these social
traits is niche expansion viaenvironmental modification (whenever
environmental modifiersare better adapted to the resulting
environmental change),suggesting that elucidating the evolutionary
functions of socialtraits has a key role in understanding microbial
ecology andbiogeography.
We stress, however, that we are not suggesting that host
rangeexpansion is necessarily the adaptive function of secretions
inpathogenic bacteria. Bacterial pathogens are often
opportunistic,and it has been recognized that many phenotypes of
importancein disease may be by-products of selection outside the
hostenvironment28. Secretions that contribute to
environmentalmodification may have evolved owing to their effects
in otherenvironments (for example, soil, vegetation and so on),
with theability to infect new hosts being a by-product or
spandrel29.While additional hosts colonized via environmental
modification
Secretome size
Gen
ome
size
Pro
babi
lity
of z
oono
sis
0 50 100 1500
1,000
2,000
3,000
4,000
5,000
6,000
7,000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
–1.5
–1
–0.5
0
0.5
1
1.5
2
Sta
ndar
dize
dre
gres
sion
coe
ffici
ent
Genome sizesecretome size
Figure 3 | Bacterial secretions increase the ability of
pathogens to infect multiple hosts. (a) Standardized regression
coefficients (multiplied by the
standard deviation of the variable) estimated by the BPMM. Dots
show the mode of the posterior distributions with lines indicating
95% CIs. Secretome
size (yellow) has a positive effect on the probability that a
pathogen is zoonotic, whereas genome size (blue) has a negative
effect. (b) Data and
BPMM predictions. Zoonoses are shown in red (n¼ 121), whereas
specialists are shown in blue (n¼ 70). Background colours indicate
the predictions ofthe BPMM.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594
4 NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
may not be of demographic significance for bacteria in all
cases,our results suggest that the secretome provides a powerful
tool toopen up new environments for bacteria to which they
canpotentially further adapt.
Our results provide strong support for the idea that
cooperativesecretions are an important driver of host range
evolution inbacteria. However, it is possible that some unmeasured
ecologicalor genomic variable that correlates with secretome and
genomesize in bacteria is the proximate driver of host range
expansion.However, our results lead to three key experimentally
testablepredictions for future work to establish the direct role
ofenvironmental modification via cooperative secretions in
deter-mining host range. First, we predict that cooperative
secretionswill simplify and standardize both the nutrient
environment andresident bacterial communities across a range of
hosts. Second,we predict that strains and/or species with the
combination of alarge secretome and small genome will show
increased ability tocolonize different hosts in the lab. Finally,
we predict that thepresence of environmentally modifying secretions
will reduce thegrowth rate of specialist pathogens (relative to
environmentalmodifier strains), giving strains that produce them an
advantageover specialists in co-infections.
Both theory and bioinformatic analyses suggest that genescoding
for social secretions are frequently associated with mobilegenetic
elements14,30. Combined with our results, this suggeststhat
monitoring for the acquisition of large numbers of secretedproteins
via horizontal gene transfer may help predict whichpathogenic
bacteria are likely to expand their host range tohumans. Given that
such monitoring of mobile genetic elementsis frequently carried out
to assess the spread of antibioticresistance genes and virulence
factors, such monitoring appearsfeasible to implement.
While our results highlight a previously unrecognized riskfactor
for host range expansion in pathogens, major challengesremain in
integrating these results with previous work on riskfactors for
zoonosis. Previous large-scale studies on risk factors forzoonosis
have largely focused on ecological and epidemiologicalfactors (for
example, wildlife diversity and human populationdensity5) governing
when and where zoonoses are likely to arise,rather than the
organismal traits that govern which species are
most likely to emerge as zoonoses5–7. Integrating these
twoperspectives on the risk factors for zoonosis will require
targetedsampling of pathogen communities across a spectrum
ofecological conditions to address how organismal traits
andepidemiological factors combine to determine host range
shifts.
Our ability to cooperatively modify and standardize
ourenvironment is commonly seen as a key element in humans’success
in colonizing virtually every terrestrial habitat on
theearth9,31,32. Our results show that this mechanism for
achievinggeneralism is not confined to humans and is widespread
acrossbacteria. Sociality appears to be just as important for the
spread ofbacterial species to new niches as it has been in human
history.
MethodsComparative analysis. We gathered data on whether 191
species of bacteria thatare pathogenic to humans are zoonotic (that
is, can naturally transmit betweenhumans and other vertebrate
hosts, n¼ 121) or azoonotic (that is, only infecthumans, n¼ 70)
from a previous collation4. For these species we also collated
theirsecretome (that is, proteins with an extracellular
localization) and genome sizesfrom PSORTdb24. We included all
available fully sequenced genomes within aspecies and their
associated plasmids and took the mean value per strain withineach
species (Supplementary Table 1 contains all data used, and a list
of genomesused is given in Supplementary Table 2). We used the
SUPERFAMILY whole-genome-based phylogeny25, which has the advantage
of minimizing the effects ofhorizontal gene transfer on the tree
topology. For each species in our analysis, weused the type strain
to produce the phylogeny.
We used a BPMM approach to test the effects of secretome and
genome sizes onthe probability that a species is zoonotic. Analyses
were implemented in R using thepackage MCMCglmm23. We fit a model
with a binomial error structure andgenome and secretome size as
predictors, and the phylogenetic covariance matrixas a random
effect. We used a weakly informative Gelman prior for
fixedeffects33,34. We specified a prior of an inverse Wishart
distribution for the randomeffect. The residual variance
(overdispersion) was fixed to 1, as this cannot beestimated with
binary data. Parameter estimates were subsequently scaled underthe
assumption that the true residual variance is 0. We ran the
analysis for3,000,000 iterations with a burn-in of 500,000 and
thinning interval of 1,000 tominimize autocorrelation in the
chains. We used the Gelman–Rubin test35,36,as well as visual
inspection of traces, on three independent chains to ensuremodel
convergence. Statistics quoted are modes and 95% CIs for the
posteriordistributions. Code for prior and model specification was
as follows: Prior o- list(B¼ list(mu¼ c(0,0,0), V¼
gelman.prior(Bsecretome_sizeþ genome_size, data¼mydata, scale¼ 1þ
1þ pi^2/3)), R¼ list(V¼ 1, fix¼ 1), G¼ list(G1¼ list(V¼
diag(1)*0.1, nu¼ 1))), Model o-
MCMCglmm(zoonoticBsecretome_sizeþgenome_size, family¼
‘‘categorical’’, data¼mydata, prior¼ Prior, pedigree¼ tree,
Clearance rate
Con
tact
rat
e be
twee
nho
st s
peci
es
Classical generalist
Strain invades
Strain fails to invade
0 0.025 0.05 0.075 0.1 0 0.025 0.05 0.075 0.10
0.02
0.04
0.06
0.08
0.1
Clearance rate
Strain invades
Strain fails to invade0
0.02
0.04
0.06
0.08
0.1Environmental modifier
Rep
rodu
ctiv
e nu
mbe
r
0
0.5
1
1.5
2
Figure 4 | Invading a population of specialists. Plotted is the
‘basic reproductive number’ (number of new infections created per
unit time when the
pathogen is rare) of classical generalists (a) and environmental
modifiers (b) when invading a population of specialists from our
epidemiological
model. The x and y axes are the rate at which infections are
cleared (a) and the contact rate between host species (bb),
respectively. High reproductivenumbers are red and low are blue.
The yellow dashed line indicates where the reproductive number
equals 1. At values above 1 the strain can invade.
Our model predicts that a strain using environmental
modification via secretions can invade a resident population of
specialist strains under a wider
range of conditions than a classical generalist strain can
(smaller area above yellow dashed line in a than in b). Here s¼
1.5, g¼ 1, b¼ 1.25, c¼0.25 and thewithin-host species contact rate,
bw¼0.1. Environmental modifiers are better able to invade a
population of specialists than classical generalists,despite
identical within-host growth rates in single strain infections, as
environmental modifiers alter the host environment to a form that
specialists are not
adapted to. This result holds whenever b4c.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594 ARTICLE
NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications 5
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
scale¼ F, nitt¼ 3000000, burnin¼ 500000, thin¼ 1000, verbose¼ F,
slice¼T,nodes¼ ‘‘TIPS’’).
Theoretical model. We use the framework of a two-host species
epidemiologicalmodel to examine the scenarios in which a
environmental modifier strategy isfavoured. We first describe our
model for the intra-host dynamics of each strain,before turning our
attention to the epidemiological dynamics.
We consider four possible strategies that a pathogen can take;
they can be aspecialist on host species H1 or H2 (S1 and S2), a
classical generalist (G) or aenvironmental modifier (EM). A
classical generalist has growth g in host species H1and H2. A
specialist has growth rate s in the host species they specialize
on. Aspecialist’s growth rate is 0 in the alternative host species.
We assume that s4g, thatis, there is a cost of generalism. The
environmental modifier strategy attempts tomodify the environment
they experience in species H1 and H2 to a common type ofenvironment
to which they are adapted. It has a frequency-dependent growth rate
ofbEM� c (where EM is the frequency of the environmental modifier
strain within thefocal host) when it infects either host species H1
or H2. The environmental modifiermodifies the current host
environment towards a modified environment to extentEM, thus
receiving a growth benefit of bEM, with a cost of c for investment
inenvironmental modification. When in competition with a
environmental modifierwithin a host a generalist will now have
growth rate g(1�EM), as the environmentalmodifier modifies the host
to species HX as it increases in frequency. Similarly, aspecialist
co-infecting its preferred host species with a environmental
modifier willhave growth rate s(1� EM).
Within-host dynamics. We use the replicator equation26 to model
the within-hostdynamics of these strains. The replicator equation
can be written as
dxidt¼ xi fiðxÞ�fðxÞð Þ; fðxÞ ¼
Xnj¼1
xjfjðxÞ ð1Þ
where xi is the proportion of individuals within the focal host
belonging to strain i,x is a vector of the frequencies of each
strain within the focal host, fi(x) is thegrowth rate of strain i
given strain frequencies x and f(x) is the mean growth rateof the
strains within the focal host.
Using our assumptions and equation (1) we can now write the
dynamics of ourthree strategies in host species Hi as
dSidt¼Si sð1� EMÞ� sSið1� EMÞþ gGð1�EMÞþEMðbEM� cÞð Þð Þ
dGdt¼G gð1� EMÞ� sSið1� EMÞþ gGð1� EMÞþEMðbEM� cÞð Þð Þ
dEMdt¼EM bEM� c� sSið1� EMÞþ gGð1�EMÞþEMðbEM� cÞð Þð Þ
ð2Þ
where EM, G, S1 and S2 are the frequencies of each strain within
the focal host. Aswe will make an assumption of superinfection in
our epidemiological model (seebelow), we need to only consider
pairwise competition between strain types withina host. Also, as
specialists have a growth rate of 0 in the alternative host, we
neednot consider this scenario.
Let us first consider competition between a specialist in its
preferred hostspecies and generalist. Setting EM¼ 0 and SiþG¼ 1 a
standard stability analysis37shows that Si*¼ 1 is the only stable
equilibrium as long as s4g, meaning that aspecialist in its
preferred host will always outcompete a classical generalist. When
aspecialist competes with a environmental modifier within a host
(that is, settingG¼ 0 and Siþ EM¼ 1) there are two possible stable
equilibria at EM*¼ 0 andEM*¼ 1, separated by a repeller at EM¼ (cþ
s)/(bþ s). If EM4(cþ s)/(bþ s) thenenvironmental modifiers sweep to
fixation and the equilibrium is EM*¼ 1, while ifEMo(cþ s)/(bþ s),
specialists win out and the equilibrium is EM*¼ 0. Similarly,when
generalists and environmental modifiers compete within a host (Si¼
0,Gþ EM¼ 1) there are two stable equilibria at EM*¼ 0 and EM*¼ 1,
separated by arepeller at EM¼ (cþ g)/(bþ g). We note that these
dynamics are similar to those ofthe classic stag-hunt game38, and a
previous analysis of immune systemprovocation by pathogens to
exclude competitors12. We can then generate thefollowing rules from
these within-host dynamics for inclusion in ourepidemiological
model:
1. Si never infects host species Hj, where jai.2. Si always
outcompetes G in host species Hi.3. EM outcompetes Si in host
species Hi with probability 1� (cþ s)/(bþ s), while
Si outcompetes EM with probability (cþ s)/(bþ s).4. EM
outcompetes G in either host species with probability 1� (cþ g)/(bþ
g),
while G outcompetes EM with probability (cþ g)/(bþ g).
Epidemiological dynamics. We use a
susceptible–infected–susceptible model forthe epidemiological
dynamics. We stress that this is the simplest possibledescription
of the epidemiological dynamics and will not hold for most
bacterialspecies, many of which will show environmental growth.
However, this modelallows us to gain some insights into the
epidemiological consequences of envir-onmental modification, while
remaining tractable. We also stress that the results ofour model
for within-host competition hold regardless of these
epidemiologicalassumptions.
We will assume that within-host dynamics occur on much faster
timescale thanthe epidemiological dynamics so that strain
replacement occurs instantaneouslyon the epidemiological timescale
and co-infection can be ignored (that is, asuperinfection model).
We can write the generic dynamics for a single strainin
susceptible–infected–susceptible model with two host species, under
theassumption that both host species show identical epidemiological
properties, as
dH1;Xdt¼ bwH1;X þ qXbbH2;X� �
1�H1;X � pH1;Y� �
� pH1;X bwH1;Y þ qYbbH2;Y� �
� aH1;XdH2;X
dt¼ bwH2;X þ qXbbH1;X� �
1�H2;X � pH2;Y� �
� pH2;X bwH2;Y þ qYbbH1;Y� �
� aH2;X
ð3Þ
where, Hi,Z is the proportion of host species i infected with
strain Z, a is theclearance rate of infections, bw is the contact
rate within a host species, bb is thecontact rate between the host
species, p is the probability that strain Y outcompetesstrain X
within a host and qZA{0,1} denotes whether strain Z can infect both
hostspecies. In each differential equation the first term captures
the spread of strain Xto new hosts of species i, the second term
captures replacement of strain X in hostspecies i by strain Y and
the third term captures the clearance of infections.
Combining this model framework with the assumptions and results
of ourwithin-host competition model we can write the
epidemiological dynamics as
dH1;Sdt¼H1;S bw 1�H1;S � 1�
cþ sbþ s
� �H1;EM
� ��
� bwH1;EM þ bbH2;EM� �
1� cþ sbþ s
� �� a�
dH2;Sdt¼H2;S bw 1�H2;S � 1�
cþ sbþ s
� �H2;EM
� ��
� bwH2;EM þ bbH1;EM� �
1� cþ sbþ s
� �� a�
dH1;Gdt¼H1;Gbw 1�H1;G �H1;S � 1�
cþ gbþ g
� �H1;EM
� �
þH2;Gbb 1�H1G �H1;S � 1�cþ gbþ g
� �H1;EM
� �
�H1;GbwH1;S �H1;G bwH1;EM þ bbH2;EM� �
1� cþ gbþ g
� ��H1;Ga
dH2;Gdt¼H2;Gbw 1�H2;G �H2;S � 1�
cþ gbþ g
� �H2;EM
� �
þH1;Gbb 1�H2;G �H2;S � 1�cþ gbþ g
� �H2;EM
� �
�H2;GbwH2;S �H2;G bwH2;EM þ bbH1;EM� �
1� cþ gbþ g
� ��H2;Ga
dH1;EMdt
¼H1;EMbw 1�H1;EM �H1;Gcþ gbþ g �H1;S
cþ sbþ s
� �
þH2;EMbb 1�H1;EM �H1;Gcþ gbþ g �H1;S
cþ sbþ s
� �
�H1;EMbwH1;Scþ sbþ s �H1;EM bwH1;G þ bbH2;G
� � cþ gbþ g �H1;EMa
dH2;EMdt
¼H2;EMbw 1�H2;EM �H2;Gcþ gbþ g �H2;S
cþ sbþ s
� �
þH1;EMbb 1�H2;EM �H2;Gcþ gbþ g �H2;S
cþ sbþ s
� �
�H2;EMbwH2;Scþ sbþ s �H2;EM bwH2;G þ bbH1;G
� � cþ gbþ g �H2;EMa
ð4Þ
where, Hi,Z is again the proportion of host species i infected
with strain type Z, andall other parameters are as above.
Invading a population of specialists. We first consider the
potential for bothclassical generalists and environmental modifiers
to invade a population of spe-cialists. First setting
H1,G¼H2,G¼H1,EM¼H2,EM¼ 0, the stable frequencies ofspecialists are
H�1;s ¼ H�2;s ¼ bw � að Þ=bw, assuming that bw4a (that is, that
spe-cialists can exist). We now consider the invasion of rare
classical generalists andenvironmental modifiers into this
population of specialists. The key epidemiolo-gical condition for
the invasion of a strain is that their ‘reproductive number’,R041
(ref. 39). We consider the scenario where the strain of interest is
introducedfrom rarity in one of the host species (which host
species is irrelevant as we assumethey have identical
epidemiological properties). We can write the conditions
forinvasion of classical generalists and environmental modifiers
as
bw þ bbð Þ 1� bw � abw� �
aþ bw bw � abw� � 41 ð5Þ
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594
6 NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
and
bw þbbð Þ 1� bw � að Þ cþ sð Þbw bþ sð Þ� �
aþ bw bw � að Þ cþ sð Þbw bþ sð Þ� � 41; ð6Þ
respectively. Here the numerators are the rate of spread of each
strain. In the caseof the classical generalists, this is simply the
sum of the within- and between-hostspecies contact rate times the
proportion of hosts not currently infected withspecialists.
However, in the case of environmental modifiers, the number
ofspecialist hosts is weighted by (cþ s)/(bþ s) as environmental
modifiers canoutcompete a specialist within a host with probability
1� (cþ s)/(bþ s). Thedenominators represent the loss of infections
by the invading strain owing toclearance of infections and
superinfection by specialists. In the case of classicalgeneralists,
this is simply the clearance rate of infection plus the within-host
speciestransmission rate times the proportion of hosts infected by
specialists. Again,however, in the case of environmental modifiers
the number of specialist hosts isweighted by (cþ s)/(bþ s) as
environmental modifiers can outcompete a specialistwithin a host
with probability 1� (cþ s)/(bþ s). These two conditions (equations
5and 6) are equivalent whenever b¼ c, while the condition for
environmentalmodifiers to invade is more easily satisfied than that
for classical generalistswhenever b4c, that is, whenever
environmental modification has a net positiveeffect on growth rate
in a single strain infection.
Invading a population of generalists. We now consider the
conditions forinvasion of a environmental modifier strain into a
population of classical gen-eralists. First setting
H1,S¼H2,S¼H1,EM¼H2,EM¼ 0, the stable frequency ofclassical
generalists is H�
1;G¼ H�
2;G¼ bw þ bb � að Þ= bw þ bbð Þ, assuming that
bwþ bb4a (that is, that classical generalists can exist). We can
now calculate theR0 for a environmental modifier strain invading
the population of classical gen-eralists as
bw þ bbð Þ 1� bw þ bb � að Þ cþ sð Þbw þbbð Þ bþ sð Þ� �
aþ bw þbbð Þ bw þbb � að Þ cþ sð Þbw þ bbð Þ bþ sð Þ� �41
ð7Þ
which simplifies to
bw þ bbð Þ b� cð Þþ a cþ gð Þa b� cð Þþ bw þ bbð Þ cþ gð Þ
41 ð8Þ
and gives the condition
b42cþ g ð9Þfor the invasion of environmental modifiers into a
population of classicalgeneralists, meaning that for sufficiently
high benefits, environmental modifierscan invade a population of
classical generalists from rarity.
A note on the problem of cheaters. The strategy of cooperative
environmentalmodification that we have examined is in principle
susceptible to cheaters that areadapted to the modified environment
that environmental modifiers create but donot invest in, and hence
do not pay a cost for its production. This problem of
howcooperation can survive in the face of such cheating has
received considerabletheoretical and empirical attention and a
number of solutions exist13. Regulatorycontrol of these traits may
be designed such that they are only expressed whencosts are limited
and/or benefits are maximized40,41. Population structure,
eitherwithin a host or among hosts, will also favour cooperation by
ensuring cooperativestrains encounter each other more frequently13.
In addition, there may befrequency dependence between cheaters and
cooperators, leading to a mix ofboth strains at equilibrium42.
Although the potential of cheaters to undermine the evolution of
cooperativetraits involved in environmental modification is an
evolutionary problem of greatinterest, it does not pose a major
obstacle for our analyses. First, the same factorsthat favour the
environmental modification strategy in our model (high benefitsand
low costs) also limit the evolutionary potential for cheating13.
Second, in ourcomparative analysis we considered the number of
genes coding for secretions thata bacteria possesses. Given that
these genes exist, it is unlikely that cheaters havepurged the
population of all cooperation. Although cheaters may prove
asignificant obstacle in the evolution of cooperative environmental
modification,this does not denigrate our result that those bacteria
that successfully evolveenvironmental modification can achieve host
generalism.
References1. Morens, D. M., Folkers, G. K. & Fauci, A. S.
The challenge of emerging and re-
emerging infectious diseases. Nature 430, 242–249 (2004).2.
Smolinski, M. S., Hamburg, M. A. & Lederberg, J. Microbial
Threats to Health:
Emergence, Detection, and Response (National Academies Press,
2003).3. Binder, S., Levitt, A. M., Sacks, J. J. & Hughes, J.
M. Emerging infectious diseases:
public health issues for the 21st century. Science 284,
1311–1313 (1999).4. Taylor, L. H., Latham, S. M. & Woolhouse,
M. E. J. Risk factors for human
disease emergence. Phil. Trans. Roy. Soc. Lond. B 356, 983–989
(2001).
5. Jones, K. E. et al. Global trends in emerging infectious
diseases. Nature 451,990–993 (2008).
6. Karesh, W. B. et al. Ecology of zoonoses: natural and
unnatural histories.Lancet 380, 1936–1945 (2012).
7. Morse, S. S. et al. Prediction and prevention of the next
pandemic zoonosis.Lancet 380, 1956–1965 (2012).
8. Futuyma, D. J. & Moreno, G. The evolution of ecological
specialization. Ann.Rev. Ecol. Sys. 19, 207–233 (1988).
9. Odling-Smee, F. J., Laland, K. N. & Feldman, M. W. Niche
Construction: TheNeglected Process in Evolution (Princeton Univ.
Press, 2003).
10. Jones, C. G., Lawton, J. H. & Shachak, M. Organisms as
ecosystem engineers.Oikos 69, 373–386.
11. Poisot, T., Bever, J. D., Nemri, A., Thrall, P. H. &
Hochberg, M. E. A conceptualframework for the evolution of
ecological specialisation. Ecol. Lett. 14, 841–851(2011).
12. Brown, S. P., Le Chat, L. & Taddei, F. Evolution of
virulence: triggering hostinflammation allows invading pathogens to
exclude competitors. Ecol. Lett. 11,44–51.
13. West, S. A., Griffin, A. S., Gardner, A. & Diggle, S. P.
Social evolution theory formicroorganisms. Nat. Rev. Microbiol. 4,
597–607 (2006).
14. Nogueira, T. et al. Horizontal gene transfer of the
secretome drives theevolution of bacterial cooperation and
virulence. Curr. Biol. 19, 1683–1691(2009).
15. Riley, M. A. & Gordon, D. M. The ecological role of
bacteriocins in bacterialcompetition. Trends Microbiol. 7, 129–133
(1999).
16. Gardner, A., West, S. A. & Buckling, A. Bacteriocins,
spite and virulence. Proc.R. Soc. Lond. B 271, 1529–1535
(2004).
17. Korgaonkar, A., Trivedi, U., Rumbaugh, K. P. & Whiteley,
M. Communitysurveillance enhances Pseudomonas aeruginosa virulence
during polymicrobialinfection. Proc. Natl Acad. Sci. USA 110,
1059–1064 (2012).
18. Jones, S. et al. The lux autoinducer regulates production of
exoenzymevirulence determinants in Erwinia carotovora and
Pseudomonas aeruginosa.EMBO J. 12, 2477–2482 (1993).
19. Dinges, M. M., Orwin, P. M. & Schlievert, P. M.
Exotoxins of Staphylococcusaureus. Clin. Microbiol. Rev. 13, 16–34
(2000).
20. Drenkard, E. & Ausubel, F. M. Pseudomonas biofilm
formation andantibiotic resistance are linked to phenotypic
variation. Nature 416, 740–743(2002).
21. Nadell, C. D. & Bassler, B. L. A fitness trade-off
between local competitionand dispersal in Vibrio cholerae biofilms.
Proc. Natl Acad. Sci. USA 108,14181–14185 (2011).
22. Moran, N. A. Microbial minimalism: genome reduction in
bacterial pathogens.Cell 108, 583–586 (2002).
23. Hadfield, J. MCMC methods for multi-response generalised
linear mixedmodels: the MCMCglmm R package. J. Stat. Softw. 33,
1–22 (2010).
24. Yu, N. Y., Laird, M. R., Spencer, C. & Brinkman, F. S.
L. PSORTdb: anexpanded, auto-updated, user-friendly protein
subcellular localisation databasefor Bacteria and Archaea. Nucleic
Acids Res. 39, D241–D244 (2011).
25. Wilson, D. et al. SUPERFAMILY–sophisticated comparative
genomics, datamining, visualization and phylogeny. Nucleic Acid
Res. 37, D380–D386 (2009).
26. Weibull, J. W. Evolutionary Game Theory (MIT Press,
1995).27. Brown, S. P. & Buckling, A. A social life for
discerning microbes. Cell 135,
600–603 (2008).28. Brown, S. P., Cornforth, D. M. & Mideo,
N. Evolution of virulence in
opportunistic pathogens: generalism, plasticity and control.
Trends Microbiol.20, 336–342 (2012).
29. Gould, S. J. & Lewtonin, R. C. The spandrels of San
Marco and the Panglossianparadigm: a critique of the adaptationist
programme. Proc. R. Soc. B 205,581–598 (1979).
30. Rankin, D. J., Rocha, E. P. & Brown, S. P. What traits
are coded on mobilegenetic elements and why? Heredity 106, 1–10
(2011).
31. Boyd, R. & Richerson, P. Culture and the Evolutionary
Process (Univ. ChicagoPress, 1985).
32. Boyd, R., Richerson, P. & Heinrich, J. The cultural
niche: Why social learning isessential for human adaptation. Proc.
Natl Acad. Sci. USA 108, 10918–10925(2011).
33. Gelman, A., Jakulin, A., Pittau, M. G. & Su, Y.-S. A
weakly informative defaultprior distribution for logistic and other
regression models. Ann. Appl. Stat. 2,1360–1383 (2008).
34. Hadfield, J. D., Heap, E. A., Bayer, F., Mittell, E. A.
& Crouch, N. M.Disentangling genetic and prenatal sources of
familial resemblance acrossontogeny in a wild passerine. Evolution
67, 2701–2713 (2013).
35. Plummer, M., Best, N., Cowles, K. & Vines, K. CODA:
convergence diagnosisand output analysis for MCMC. R News 6, 1–25
(2006).
36. Gelman, A. & Rubin, D. B. Inference from iterative
simulation using multiplesequences. Stat. Sci. 7, 457–472
(1992).
37. Otto, S. P. & Day, T. A Biologist’s Guide to
Mathematical Modelling in Ecologyand Evolution (Princeton Univ.
Press, 2007).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594 ARTICLE
NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications 7
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
38. Skyrms, B. The Stag Hunt and the Evolution of Social
Structure (CambridgeUniv. Press, 2004).
39. Anderson, R. M. & May, R. M. Infectious Diseases of
Humans: Dynamics andControl (Oxford Univ. Press, 1991).
40. Xavier, J. B., Kim, W. & Foster, K. A molecular
mechanism that stabilizescooperative secretions in Pseudomonas
aeruginosa. Mol. Microbiol. 79,166–179 (2011).
41. Darch, S. E., West, S. A., Winzer, S. A. & Diggle, S. P.
Density-dependent fitnessbenefits of quorum-sensing bacterial
populations. Proc. Natl Acad. Sci. USA109, 8259–8263 (2012).
42. Ross-Gillespie, A., Gardner, A., West, S. A. & Griffin,
A. S. Frequencydependence and cooperation: theory and a test with
bacteria. Am. Nat. 170,331–342 (2007).
AcknowledgementsWe thank Mark Woolhouse, Ally Phillimore, Rolf
Kümmerli, Pedro Vale, Roman Popat,Daniel Cornforth, Richard Allen,
Adam Kane and Andrew Jackson for helpful discus-sions and/or
comments on previous versions of this manuscript. L.M. was
supported by aResearch Fellowship as part of a Wellcome Trust
Strategic grant to the Centre forImmunity, Infection and Evolution
(grant reference number 095831). M.V. was sup-ported by a Newton
International Fellowship from the Royal Society. S.P.B. was
fundedby the EPSRC (EP/H032436/1).
Author contributionsL.M. and S.P.B. developed the theoretical
model. L.M. and M.V. collated and analysed thedata. L.M. drafted
the manuscript. All authors contributed to conceptual
development,study design and manuscript revision.
Additional informationSupplementary Information accompanies this
paper at http://www.nature.com/naturecommunications
Competing financial interests: The authors declare no competing
financial interests.
Reprints and permission information is available online at
http://npg.nature.com/reprintsandpermissions/
How to cite this article: McNally, L. et al. Cooperative
secretions facilitate host rangeexpansion in bacteria. Nat. Commun.
5:4594 doi: 10.1038/ncomms5594 (2014).
This work is licensed under a Creative Commons Attribution
4.0International License. The images or other third party material
in this
article are included in the article’s Creative Commons license,
unless indicated otherwisein the credit line; if the material is
not included under the Creative Commons license,users will need to
obtain permission from the license holder to reproduce the
material.To view a copy of this license, visit
http://creativecommons.org/licenses/by/4.0/
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms5594
8 NATURE COMMUNICATIONS | 5:4594 | DOI: 10.1038/ncomms5594 |
www.nature.com/naturecommunications
& 2014 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunicationshttp://www.nature.com/naturecommunicationshttp://npg.nature.com/reprintsandpermissions/http://npg.nature.com/reprintsandpermissions/http://creativecommons.org/licenses/by/4.0/http://www.nature.com/naturecommunications
title_linkResultsComparative analysis
Figure™1Environment-modifying secretions as a route to host
generalism.We consider a scenario where pathogens can potentially
transmit both within and among host species. Whereas specialists
match their hosts closely (matching colours), generalists that
iTheoretical model
Figure™2The phylogenetic distribution of zoonosis, genome size
and secretome size.The phylogenetic distribution of zoonotic
status, secretome size and genome size is shown. Large genomes and
secretomes are those greater than the median and small are less
DiscussionFigure™3Bacterial secretions increase the ability of
pathogens to infect multiple hosts.(a) Standardized regression
coefficients (multiplied by the standard deviation of the variable)
estimated by the BPMM. Dots show the mode of the posterior
distributionMethodsComparative analysis
Figure™4Invading a population of specialists.Plotted is the
’basic reproductive numberCloseCurlyQuote (number of new infections
created per unit time when the pathogen is rare) of classical
generalists (a) and environmental modifiers (b) when invading a
pTheoretical modelWithin-host dynamicsEpidemiological
dynamicsInvading a population of specialistsInvading a population
of generalistsA note on the problem of cheaters
MorensD. M.FolkersG. K.FauciA. S.The challenge of emerging and
re-emerging infectious diseasesNature4302422492004SmolinskiM.
S.HamburgM. A.LederbergJ.Microbial Threats to Health: Emergence,
Detection, and ResponseNational Academies Press2003BinderS.LevittWe
thank Mark Woolhouse, Ally Phillimore, Rolf Kümmerli, Pedro Vale,
Roman Popat, Daniel Cornforth, Richard Allen, Adam Kane and Andrew
Jackson for helpful discussions andsolor comments on previous
versions of this manuscript. L.M. was supported by a
ReseACKNOWLEDGEMENTSAuthor contributionsAdditional information